added simplified Unicode support for non-locale-enabled builds

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

6 files changed, 331 insertions, 24 deletions

diff --git a/Config.in b/Config.in
index ee706eb..99f814e 100644
--- a/Config.in
+++ b/Config.in
@@ -30,18 +30,6 @@ config EXTRA_COMPAT
 	  some GNU extensions in libc. You probably only need this option
 	  if you plan to run busybox on desktop.
 
-config FEATURE_ASSUME_UNICODE
-	bool "Assume that 1:1 char/glyph correspondence is not true"
-	default n
-	help
-	  This makes various applets aware that one byte is not
-	  one character on screen.
-
-	  Busybox aims to eventually work correctly with Unicode displays.
-	  Any older encodings are not guaranteed to work.
-	  Probably by the time when busybox will be fully Unicode-clean,
-	  other encodings will be mainly of historic interest.
-
 choice
 	prompt "Buffer allocation policy"
 	default FEATURE_BUFFERS_USE_MALLOC
@@ -114,6 +102,28 @@ config LOCALE_SUPPORT
 	  Enable this if your system has locale support and you would like
 	  busybox to support locale settings.
 
+config FEATURE_ASSUME_UNICODE
+	bool "Support Unicode"
+	default n
+	help
+	  This makes various applets aware that one byte is not
+	  one character on screen.
+
+	  Busybox aims to eventually work correctly with Unicode displays.
+	  Any older encodings are not guaranteed to work.
+	  Probably by the time when busybox will be fully Unicode-clean,
+	  other encodings will be mainly of historic interest.
+
+config FEATURE_CHECK_UNICODE_IN_ENV
+	bool "Check $LANG environment variable"
+	default y
+	depends on FEATURE_ASSUME_UNICODE && !LOCALE_SUPPORT
+	help
+	  With this option on, Unicode support is activated
+	  only if LANG variable has the value of the form "xxxx.utf8"
+
+	  Otherwise, Unicode support will be always enabled and active.
+
 config LONG_OPTS
 	bool "Support for --long-options"
 	default y
diff --git a/coreutils/ls.c b/coreutils/ls.c
index 8a6faf2..20b979d 100644
--- a/coreutils/ls.c
+++ b/coreutils/ls.c
@@ -30,12 +30,9 @@
  * [2009-03]
  * ls sorts listing now, and supports almost all options.
  */
-
 #include "libbb.h"
+#include "unicode.h"
 
-#if ENABLE_FEATURE_ASSUME_UNICODE
-#include <wchar.h>
-#endif
 
 /* This is a NOEXEC applet. Be very careful! */
 
@@ -296,9 +293,8 @@ enum {
 /* libbb candidate */
 static size_t mbstrlen(const char *string)
 {
-	size_t width = mbsrtowcs(NULL /*dest*/, &string,
-				MAXINT(size_t) /*len*/, NULL /*state*/);
-	if (width == (size_t)-1)
+	size_t width = mbstowcs(NULL, string, INT_MAX);
+	if (width == (size_t)-1L)
 		return strlen(string);
 	return width;
 }
@@ -932,6 +928,8 @@ int ls_main(int argc UNUSED_PARAM, char **argv)
 
 	INIT_G();
 
+	check_unicode_in_env();
+
 	all_fmt = LIST_SHORT |
 		(ENABLE_FEATURE_LS_SORTFILES * (SORT_NAME | SORT_FORWARD));
 
diff --git a/include/unicode.h b/include/unicode.h
new file mode 100644
index 0000000..be64a50
--- a/dev/null
+++ b/include/unicode.h
@@ -0,0 +1,57 @@
+/* vi: set sw=4 ts=4: */
+/*
+ * Licensed under the GPL version 2, see the file LICENSE in this tarball.
+ */
+#ifndef UNICODE_H
+#define UNICODE_H 1
+
+#if !ENABLE_FEATURE_ASSUME_UNICODE
+
+# define check_unicode_in_env() ((void)0)
+
+#else
+
+# if ENABLE_LOCALE_SUPPORT
+
+#  include <wchar.h>
+#  include <wctype.h>
+#  define check_unicode_in_env() ((void)0)
+
+# else
+
+#  if !ENABLE_FEATURE_CHECK_UNICODE_IN_ENV
+#   define check_unicode_in_env() ((void)0)
+#  else
+void check_unicode_in_env(void) FAST_FUNC;
+#  endif
+
+#  undef MB_CUR_MAX
+#  define MB_CUR_MAX 6
+
+/* Prevent name collisions */
+#  define wint_t    bb_wint_t
+#  define mbstate_t bb_mbstate_t
+#  define mbstowcs  bb_mbstowcs
+#  define wcstombs  bb_wcstombs
+#  define wcrtomb   bb_wcrtomb
+#  define iswspace  bb_iswspace
+#  define iswalnum  bb_iswalnum
+#  define iswpunct  bb_iswpunct
+
+typedef int32_t wint_t;
+typedef struct {
+	char bogus;
+} mbstate_t;
+
+size_t mbstowcs(wchar_t *dest, const char *src, size_t n) FAST_FUNC;
+size_t wcstombs(char *dest, const wchar_t *src, size_t n) FAST_FUNC;
+size_t wcrtomb(char *s, wchar_t wc, mbstate_t *ps) FAST_FUNC;
+int iswspace(wint_t wc) FAST_FUNC;
+int iswalnum(wint_t wc) FAST_FUNC;
+int iswpunct(wint_t wc) FAST_FUNC;
+
+# endif
+
+#endif
+
+#endif
diff --git a/libbb/Kbuild b/libbb/Kbuild
index 70dc48d..efd04e3 100644
--- a/libbb/Kbuild
+++ b/libbb/Kbuild
@@ -139,6 +139,8 @@ lib-$(CONFIG_HWCLOCK) += rtc.o
 lib-$(CONFIG_RTCWAKE) += rtc.o
 lib-$(CONFIG_FEATURE_CHECK_NAMES) += die_if_bad_username.o
 
+lib-$(CONFIG_FEATURE_ASSUME_UNICODE) += unicode.o
+
 # We shouldn't build xregcomp.c if we don't need it - this ensures we don't
 # require regex.h to be in the include dir even if we don't need it thereby
 # allowing us to build busybox even if uclibc regex support is disabled.
diff --git a/libbb/lineedit.c b/libbb/lineedit.c
index e5d0c1b..ab32972 100644
--- a/libbb/lineedit.c
+++ b/libbb/lineedit.c
@@ -34,10 +34,7 @@
  * PS1='\[\033[01;32m\]\u@\h\[\033[01;34m\] \w \$\[\033[00m\] '
  */
 #include "libbb.h"
-#if ENABLE_FEATURE_ASSUME_UNICODE
-# include <wchar.h>
-# include <wctype.h>
-#endif
+#include "unicode.h"
 
 /* FIXME: obsolete CONFIG item? */
 #define ENABLE_FEATURE_NONPRINTABLE_INVERSE_PUT 0
@@ -1581,7 +1578,7 @@ static int lineedit_read_key(char *read_key_buffer)
 				return ic;
 			unicode_buf[unicode_idx++] = ic;
 			unicode_buf[unicode_idx] = '\0';
-			if (mbstowcs(&wc, unicode_buf, 1) < 1 && unicode_idx < MB_CUR_MAX) {
+			if (mbstowcs(&wc, unicode_buf, 1) != 1 && unicode_idx < MB_CUR_MAX) {
 				delay = 50;
 				goto poll_again;
 			}
@@ -1636,6 +1633,8 @@ int FAST_FUNC read_line_input(const char *prompt, char *command, int maxsize, li
 		return len;
 	}
 
+	check_unicode_in_env();
+
 // FIXME: audit & improve this
 	if (maxsize > MAX_LINELEN)
 		maxsize = MAX_LINELEN;
diff --git a/libbb/unicode.c b/libbb/unicode.c
new file mode 100644
index 0000000..a99f5ed
--- a/dev/null
+++ b/libbb/unicode.c
@@ -0,0 +1,241 @@
+/* vi: set sw=4 ts=4: */
+/*
+ * Unicode support routines.
+ *
+ * Copyright (C) 2008 Denys Vlasenko
+ *
+ * Licensed under GPL version 2, see file LICENSE in this tarball for details.
+ */
+#include "libbb.h"
+
+/* if LOCALE_SUPPORT, libc locale stuff takes care of it, else: */
+
+#if !ENABLE_LOCALE_SUPPORT
+#include "unicode.h"
+
+/* 0: not known yet,
+ * 1: not unicode (IOW: assuming one char == one byte)
+ * 2: unicode
+ */
+# if !ENABLE_FEATURE_CHECK_UNICODE_IN_ENV
+#  define unicode_is_enabled 2
+# else
+static smallint unicode_is_enabled;
+void FAST_FUNC check_unicode_in_env(void)
+{
+	char *lang;
+
+	if (unicode_is_enabled)
+		return;
+	unicode_is_enabled = 1;
+
+	lang = getenv("LANG");
+	if (!lang || !strstr(lang, ".utf8"))
+		return;
+
+	unicode_is_enabled = 2;
+}
+# endif
+
+static size_t wcrtomb_internal(char *s, wchar_t wc)
+{
+	uint32_t v = wc;
+
+	if (v <= 0x7f) {
+		*s = v;
+		return 1;
+	}
+
+	/* 80-7FF -> 110yyyxx 10xxxxxx */
+	if (v <= 0x7ff) {
+		s[1] = (v & 0x3f) | 0x80;
+		v >>= 6;
+		s[0] = v | 0xc0;
+		return 2;
+	}
+
+	/* 800-FFFF -> 1110yyyy 10yyyyxx 10xxxxxx */
+	if (v <= 0xffff) {
+		s[2] = (v & 0x3f) | 0x80;
+		v >>= 6;
+		s[1] = (v & 0x3f) | 0x80;
+		v >>= 6;
+		s[0] = v | 0xe0;
+		return 3;
+	}
+
+	/* RFC 3629 says that Unicode ends at 10FFFF */
+
+	/* 10000-1FFFFF -> 11110zzz 10zzyyyy 10yyyyxx 10xxxxxx */
+	if (v <= 0x1fffff) {
+		s[3] = (v & 0x3f) | 0x80;
+		v >>= 6;
+		s[2] = (v & 0x3f) | 0x80;
+		v >>= 6;
+		s[1] = (v & 0x3f) | 0x80;
+		v >>= 6;
+		s[0] = v | 0xf0;
+		return 4;
+	}
+
+	/* 200000-3FFFFFF -> 111110tt 10zzzzzz 10zzyyyy 10yyyyxx 10xxxxxx */
+	if (v <= 0x3ffffff) {
+		s[4] = (v & 0x3f) | 0x80;
+		v >>= 6;
+		s[3] = (v & 0x3f) | 0x80;
+		v >>= 6;
+		s[2] = (v & 0x3f) | 0x80;
+		v >>= 6;
+		s[1] = (v & 0x3f) | 0x80;
+		v >>= 6;
+		s[0] = v | 0xf8;
+		return 5;
+	}
+
+	/* 4000000-FFFFFFFF -> 111111tt 10tttttt 10zzzzzz 10zzyyyy 10yyyyxx 10xxxxxx */
+	s[5] = (v & 0x3f) | 0x80;
+	v >>= 6;
+	s[4] = (v & 0x3f) | 0x80;
+	v >>= 6;
+	s[3] = (v & 0x3f) | 0x80;
+	v >>= 6;
+	s[2] = (v & 0x3f) | 0x80;
+	v >>= 6;
+	s[1] = (v & 0x3f) | 0x80;
+	v >>= 6;
+	s[0] = v | 0xfc;
+	return 6;
+}
+
+size_t FAST_FUNC wcrtomb(char *s, wchar_t wc, mbstate_t *ps UNUSED_PARAM)
+{
+	if (unicode_is_enabled != 2) {
+		*s = wc;
+		return 1;
+	}
+
+	return wcrtomb_internal(s, wc);
+}
+
+size_t FAST_FUNC wcstombs(char *dest, const wchar_t *src, size_t n)
+{
+	size_t org_n = n;
+
+	if (unicode_is_enabled != 2) {
+		while (n) {
+			wchar_t c = *src++;
+			*dest++ = c;
+			if (c == 0)
+				break;
+			n--;
+		}
+		return org_n - n;
+	}
+
+	while (n >= MB_CUR_MAX) {
+		wchar_t wc = *src++;
+		size_t len = wcrtomb_internal(dest, wc);
+
+		if (wc == L'\0')
+			return org_n - n;
+		dest += len;
+		n -= len;
+	}
+	while (n) {
+		char tbuf[MB_CUR_MAX];
+		wchar_t wc = *src++;
+		size_t len = wcrtomb_internal(tbuf, wc);
+
+		if (len > n)
+			len = n;
+		memcpy(dest, tbuf, len);
+		if (wc == L'\0')
+			return org_n - n;
+		dest += len;
+		n -= len;
+	}
+	return org_n - n;
+}
+
+size_t FAST_FUNC mbstowcs(wchar_t *dest, const char *src, size_t n)
+{
+	size_t org_n = n;
+
+	if (unicode_is_enabled != 2) {
+		while (n) {
+			unsigned char c = *src++;
+			*dest++ = c;
+			if (c == 0)
+				break;
+			n--;
+		}
+		return org_n - n;
+	}
+
+	while (n) {
+		int bytes;
+		unsigned c = (unsigned char) *src++;
+
+		if (c <= 0x7f) {
+			*dest++ = c;
+			if (c == '\0')
+				break;
+			n--;
+			continue;
+		}
+
+		/* 80-7FF -> 110yyyxx 10xxxxxx */
+		/* 800-FFFF -> 1110yyyy 10yyyyxx 10xxxxxx */
+		/* 10000-1FFFFF -> 11110zzz 10zzyyyy 10yyyyxx 10xxxxxx */
+		/* 200000-3FFFFFF -> 111110tt 10zzzzzz 10zzyyyy 10yyyyxx 10xxxxxx */
+		/* 4000000-FFFFFFFF -> 111111tt 10tttttt 10zzzzzz 10zzyyyy 10yyyyxx 10xxxxxx */
+		bytes = 0;
+		do {
+			c <<= 1;
+			bytes++;
+		} while ((c & 0x80) && bytes < 6);
+		if (bytes == 1)
+			return (size_t) -1L;
+		c = (uint8_t)(c) >> bytes;
+
+		while (--bytes) {
+			unsigned ch = (unsigned char) *src++;
+			if ((ch & 0xc0) != 0x80) {
+				return (size_t) -1L;
+			}
+			c = (c << 6) + (ch & 0x3f);
+		}
+
+		/* TODO */
+		/* Need to check that c isn't produced by overlong encoding */
+		/* Example: 11000000 10000000 converts to NUL */
+		/* 11110000 10000000 10000100 10000000 converts to 0x100 */
+		/* correct encoding: 11000100 10000000 */
+		if (c <= 0x7f) { /* crude check */
+			return (size_t) -1L;
+			//or maybe: c = 0xfffd; /* replacement character */
+		}
+
+		*dest++ = c;
+		n--;
+	}
+
+	return org_n - n;
+}
+
+int FAST_FUNC iswspace(wint_t wc)
+{
+	return (unsigned)wc <= 0x7f && isspace(wc);
+}
+
+int FAST_FUNC iswalnum(wint_t wc)
+{
+	return (unsigned)wc <= 0x7f && isalnum(wc);
+}
+
+int FAST_FUNC iswpunct(wint_t wc)
+{
+	return (unsigned)wc <= 0x7f && ispunct(wc);
+}
+
+#endif